<p>The LPBF-prepared Ti6Al4V alloy pieces are extremely sensitive to processing variables. In order to maintain an appropriate melt pool size and create a dense microstructure, a sufficient volume energy density has been found after a significant amount of processing area has been investigated. Due to the high dimensional characteristics of the operation space, even if the energy input density is limited, it also corresponds to a large number of processing parameter combinations. To explore the appropriate parameter combination under the suitable volume energy density, we employed the verified high-fidelity numerical model to calculate the melt pool morphology and spatter probability of the titanium alloy LPBF process, and statistics the porosity and roughness. The simulated findings serve as the foundation for the basic data set. The processing parameter space is explored using multi-objective Bayesian optimization, and experiments are conducted to confirm the prediction's accuracy. Significant variations in surface roughness, pore ratio generation, and spatter behavior were also noted in various parameter combinations with a constant volumetric energy density (50 J/mm3), according to this study. In addition to effectively identifying ideal operating parameters, Bayesian optimization reveals the nonlinear correlations between laser power, scan speed, and overlapping ratio. In comparison to empirical settings, the optimized parameter set (195 W, 1.0 m/s, 0.45), which also enhanced tensile strength and elongation about 6%, resulted in a 43.55% decrease in surface roughness. The observed trends in the BO-predicted window show that spatter and porosity exhibit opposing responses to parameter variation, highlighting the need for balanced trade-offs in LPBF parameter design.</p> Graphical Abstract <p></p>